From 6d6b3504ac3a0c6d62c7759a9b91ae6372227f92 Mon Sep 17 00:00:00 2001
From: Alan Jean <afinegameofnil@gmail.com>
Date: Mon, 30 Jan 2023 23:15:56 +0400
Subject: [PATCH] =?UTF-8?q?feat(inference):=20add=20compose=20file=20for?=
 =?UTF-8?q?=20the=20inference=20stack=20=F0=9F=90=B3?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Initial inference stack definition.
---
 docker-compose.inference.yaml | 63 +++++++++++++++++++++++++++++++++++
 1 file changed, 63 insertions(+)
 create mode 100644 docker-compose.inference.yaml

diff --git a/docker-compose.inference.yaml b/docker-compose.inference.yaml
new file mode 100644
index 00000000..38609b36
--- /dev/null
+++ b/docker-compose.inference.yaml
@@ -0,0 +1,63 @@
+version: "3.7"
+
+services:
+  inference-server:
+    build:
+      dockerfile: docker/inference/Dockerfile.server
+      context: .
+    image: oasst-inference-server
+    environment:
+      - "PORT=8000"
+      - "REDIS_HOST=redis"
+    volumes:
+      - "./oasst-shared:/opt/inference/lib/oasst-shared"
+      - "./inference/server:/opt/inference/server"
+    restart: unless-stopped
+    depends_on:
+      redis:
+        condition: service_healthy
+
+  inference-worker:
+    build:
+      dockerfile: docker/inference/Dockerfile.worker
+      context: .
+    image: oasst-inference-worker
+    environment:
+      - "BACKEND_URL=ws://inference-server:8000"
+      - "INFERENCE_SERVER_URL=http://inference-text-generation-server"
+    volumes:
+      - "./oasst-shared:/opt/inference/lib/oasst-shared"
+      - "./inference/worker:/opt/inference/worker"
+    depends_on:
+      - inference-server
+    deploy:
+      replicas: 1
+
+  inference-text-client:
+    build:
+      dockerfile: docker/inference/Dockerfile.text-client
+      context: .
+    image: oasst-inference-text-client
+    environment:
+      - "BACKEND_URL=http://inference-server:8000"
+    tty: true
+    stdin_open: true
+    volumes:
+      - "./inference/text-client:/opt/inference/text-client"
+    restart: unless-stopped
+    depends_on:
+      - inference-server
+
+  inference-text-generation-server:
+    image: ykilcher/text-generation-inference
+    environment:
+      - "MODEL_ID=distilgpt2"
+
+  redis:
+    image: redis
+    restart: always
+    healthcheck:
+      test: ["CMD-SHELL", "redis-cli ping | grep PONG"]
+      interval: 2s
+      timeout: 2s
+      retries: 10